[Issue 9438] Strange RefCounted stack overflow

[d-bugmail at puremagic.com]

http://d.puremagic.com/issues/show_bug.cgi?id=9438


monarchdodra at gmail.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-code
           Severity|major                       |critical


--- Comment #3 from monarchdodra at gmail.com 2013-02-01 09:09:48 PST ---
I wanted to try to get RefCounted out of the way, and I was able to get a
massively reduced test case. That said, the result is mind bogglingly wtf...

FYI: win32 on win7.64

//----
import std.stdio, std.exception;

//S is merely a struct with a pointer. And a destructor.
struct S
{
    int* _payload = null;

    ~this()
    {
        writeln("~this: ", _payload);
        if (!_payload) return;
        writeln("Should not be here...");
    }
}

void foo(int* p)
{
  throw new Exception("bla");
}

void main()
{
    int* p = S()._payload;
    writeln("wait for it...");
    foo(S()._payload);
}
//----
~this: null
wait for it...
~this: FFFF05E8
Should not be here...
object.Exception at main.d(17): bla
//----

Apparently, we create a temporary, an exception is thrown, the temporary gets
corrupted, and then things start getting awry in the destructor.

I'm getting this also for as far back as 2.55 (didn't try anything earlier), so
it doesn't seem to be a regression.

In any case, exceptions silently corrupting stack temporaries? That's a
critical in my book.

Who was it again that said our destructors weren't very well tested?

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------